Those errors were fixed in a second version of the terrorism report, released June 22. But six Democratic senators, suggesting that the administration was manipulating terrorism statistics for election-year political gain, asked Secretary of State Colin L. Powell to find out what had gone wrong, prompting the investigation by the inspector general. A copy of the inspector general's conclusions, marked "sensitive but unclassified", was obtained by The Times.
The annual report has been mandated by Congress since 1987 as the government's primary reference tool on worldwide terrorist activity, trends and groups and the U.S. response to them.
The investigators, overseen by the State Department's acting inspector general, Cameron R. Hume, stopped short of calling for a second revision of the widely circulated report. But they concluded that the report, even in its revised form, "cannot be viewed as reliable" because of the questionable statistics on terrorist attacks, casualties and other issues. The report urged better oversight and management of the annual terrorism report card.
A State Department spokesman declined to comment publicly on the internal report, but said the department had no plans to review or reissue the 2003 "Patterns" document a second time. The official said the State Department already was moving to overhaul the way it compiled terrorism statistics.
"We think it's best to just move on, and make sure we fix what needs to be fixed," the official said.
On Friday, Sen. Patrick J. Leahy (D-Vt.), one of the lawmakers who requested the investigation, said the lack of objective benchmarks to measure terrorist activity jeopardized the campaign against terrorism.
"Either through indifference or incompetence … these errors have damaged the credibility of this important assessment, undermining our ability to determine what policies and programs are effective in fighting terrorism," Leahy said.
A senior congressional official said the inspector general's findings confirmed what experts had been saying for years — that the annual "Patterns" report was seriously flawed as a tool to measure progress in the war on terrorism or analyze the rapidly changing nature of terrorism.
"We become the laughingstock if we redo it. But [not doing it] poses a serious credibility problem," said the official, a terrorism analyst on Capitol Hill. "This determines where we put our resources, what we tell other countries, what we think the trends are. And this just ruins our credibility. People just don't trust us anymore."
Michael Kraft, a senior counter-terrorism official in the State Department until earlier this year, defended the annual "Patterns" report as immensely valuable, and said it was almost impossible to be entirely accurate given all of the variables that went into analyzing terrorism.
"It's not always easy. The numbers themselves don't always mean a great deal. They have to be put in context," Kraft said. "Even with the best of efforts — and a lot of time and work goes into it — there is always going to be a certain amount of fuzziness."
In their internal report, State Department investigators pinned much of the blame for the 2003 problems on the transition to the government's new interagency Terrorist Threat Integration Center, which last year took over from the CIA responsibility for the terrorism database used to compile statistics for the annual report.
During the transition, investigators concluded, there were gaps in data entry, inadequate oversight, staff shortages and personnel changes that resulted in a lack of trained, long-term employees. In particular, the CIA-based manager of the database unit left in December 2003, but was not replaced until April 2004.
Kraft described many of the problems as "teething pains" inevitable in a new agency.
But investigators identified more systemic shortcomings, particularly a long-standing failure by the State Department, CIA and other agencies to use consistent standards to identify and classify terrorism-related events.
For example, some multiple bombings in the same city — such as bomb attacks March 25, 2003, on four U.N. police stations in Pristina, Serbia, and attacks on two embassies in Caracas, Venezuela, on Feb. 25 — were counted as single terrorism incidents.
But grenade attacks on two targets in Kashmir on April 12, and bomb attacks on two synagogues in Istanbul on Nov. 15, were each listed as two terrorism incidents.
Additionally, some items were included or dropped without apparent reason. The discovery of an explosive device at an IBM facility in Italy on March 31 was deleted without explanation from the second version of the 2003 report. But a parcel bomb hidden in a book that was sent to the Greek Consulate in Madrid on Sept. 8 was added to the revised version.
Investigators said no records or minutes were kept to explain how these decisions were made. Thus officials "could only speculate on why some events were included or not included," according to the report.
Meanwhile, there appear to be obvious inconsistencies within the revised 2003 report, said another congressional staff member.
Among them: the corrected report lists 2,738 people as casualties of international terrorist attacks in 2002 in one section, but 3,072 casualties in a separate statistical review.
Long-standing guidelines, meanwhile, have not kept pace with changes in terrorism.
The report considers international terrorism to be violence against noncombatant targets by subnational groups or clandestine agents and which involves citizens of two or more countries.
This effectively omitted countless incidents in Chechnya, and more recently in Iraq, even though the Bush administration contended that Iraq was now at the heart of the terrorism war.